[daip] IRING q?

[Eric Greisen]

Lawrence Rudnick wrote:
> So it's a calculation of error in the mean? (assuming nnz(k) is the 
> number of independent beams in ring k), as opposed to the rms scatter 
> within the ring?

No - it is a simple count of all non-zero, non-blanked pixels.  There 
has not been a correction for correlation between adjacent pixels.  That 
correlation would be tricky to determine if the rings have substantial 
width.  Therefore, these error bars are really too small by e.g a factor 
of sqrt(3) or 2 or ??  Maybe I should change this computation except 
that I think we do not do such corrections elsewhere either.

Eric